Big Idea

The argument in one line.

Most people get zero ROI from AI because they point it at demos instead of bottlenecks -- the business chain audit is the only step that makes every AI purchase worth more than it costs.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

A solo founder or small team owner who has experimented with AI tools but has not seen a revenue impact yet.
A service business owner drowning in delivery who suspects AI could handle more of the work than they currently trust it to.
A founder evaluating which part of their pipeline to automate first and looking for a prioritization framework.
Someone considering hiring but wanting to know whether AI can substitute for headcount at their current scale.

SKIP IF…

You already have a disciplined bottleneck audit process and are looking for deep technical AI implementation guides.
You want tool-agnostic advice -- this video promotes several Martell Ventures portfolio products throughout.

TL;DR

The full version, fast.

Stop buying AI tools before you know where your business is actually blocked. Every business runs on the same chain -- Attention, Leads, Sales, Delivery, Retention -- and the one stage that would most increase revenue if 10x-ed is your only target. Once you know the bottleneck, AI covers the full sales pipeline and delivery without a capacity ceiling. The 10-80-10 rule and the camcorder method give you a repeatable process to hand any task to AI while keeping your judgment at the edges.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →

Chapters

Where the time goes.

00:00 – 00:57

01 · Stop using AI like a toy

Pattern interrupt and credibility. AI should do real work, not demos.

00:57 – 03:25

02 · Give AI a real target

Business chain audit. 10x test to identify the bottleneck.

03:26 – 06:35

03 · AI generates revenue: 5-stage sales engine

Prospect, qualify, pitch, follow-up, upsell -- all delegated to AI.

06:35 – 07:50

04 · AI delivers

AI has no capacity ceiling. Young Guns internship story.

07:50 – 09:34

05 · The 10-80-10 rule

Human owns ideation and integration; AI owns execution.

09:34 – 11:30

06 · Replace yourself before AI does

AI hackathon story. Claude Code mandatory. Scaling equals becoming less necessary.

11:30 – 13:58

07 · The camcorder method

Four-step process to delegate any task to AI. Urgency close: 18-month window.

13:58 – 14:10

08 · CTA

DM YouTube OS on Instagram for the AI Company OS playbook.

Atomic Insights

Lines worth screenshotting.

The bottleneck test: if you 10x this stage tomorrow, does revenue go up or do you just create a new bottleneck?
AI does not feel rejection -- it will follow up on 20,000 cold leads with the same energy as the first one.
Most businesses are blocked at attention or leads, not delivery -- fixing delivery with AI before fixing leads is building the wrong wall.
80% done by AI is 100% better than 0% done because you were too busy.
The 10-80-10 rule: you set the definition of done, AI executes, you add the human touch -- your fingerprint stays on the work.
Telling AI to check its own work before returning results cuts correction loops in half.
Scaling a business correctly means making yourself less necessary every quarter -- guilt about that is a signal you are doing it right.
The camcorder method turns any task into an AI-ready SOP in four steps: record yourself, transcribe, feed to AI, iterate.
An AI sales call handling high-ticket subscriptions is not a future concept -- it is running in portfolio companies today.
Giving new hires AI access compresses years of on-the-job learning into weeks.
The competitive window for first movers closes in roughly 18 months -- late adopters will not compete, they will be lapped.
Rich people use AI to buy back time, then redeploy that time on things only they can do -- the leverage is not in the tool, it is in the redirect.

Takeaway

Find the bottleneck before buying any AI tool.

THE FRAMEWORK

Every AI investment that does not target your actual revenue constraint is wasted -- the business chain audit is the only prerequisite that makes everything else compound.

Run the 10x test on each stage of your pipeline (Attention, Leads, Sales, Delivery, Retention) and only invest AI resources where the answer is that revenue would go up.
AI does not feel rejection, burn out, or hit capacity, which makes it structurally better than humans at high-volume repetitive tasks like cold outreach and follow-up sequences.
The 10-80-10 rule protects quality: own the first 10% (what done looks like) and the last 10% (review and judgment), and let AI own everything in between.
The camcorder method turns any tacit process into an AI-ready SOP without writing documentation -- record yourself doing it once, transcribe, feed to the model, iterate.
Proactively replacing yourself with AI is a strategic choice; waiting means a competitor does it to your market first.
Asking AI to verify its own work against the definition of done before returning results reduces correction cycles and is worth building into every workflow.
The strongest signal you are scaling correctly is feeling less necessary over time -- resistance to that feeling is the bottleneck most founders never audit.

Glossary

Terms worth knowing.

Business chain: The five-stage revenue pipeline every business runs on: Attention, Leads, Sales, Delivery, and Retention. Auditing each stage reveals where AI investment produces the most leverage.
Bottleneck: The single stage in the business chain that, if improved, would most increase revenue. Identifying it before buying any AI tool is the central thesis of the video.
10-80-10 rule: A delegation framework where the human owns the first 10% (ideation and definition of done) and the final 10% (review and integration), while AI handles the middle 80% (execution).
Camcorder method: A four-step process for delegating tasks to AI: record yourself doing the task on screen with narration, transcribe the recording, feed it to AI, then iterate until AI can replicate 80% of it.
Definition of done: The explicit output criteria negotiated with AI during the ideation phase, used to keep execution on target before handing the task over.

Resources

Things they pointed at.

03:39toolClaude Code ↗

04:19toolManus ↗

04:58toolYour Atlas ↗

04:18toolRevio / GetRevio ↗

06:29toolLatch ↗

00:37toolApex ↗

00:00bookBuy Back Your Time (book) ↗

Quotables

Lines you could clip.

00:32

“Rich people put AI to real work, real tasks, real outputs, real problems, big stuff.”

Punchy contrast against the AI-as-toy frame; standalone without context→ TikTok hook↗ Tweet quote

00:57

“Before you buy a single tool, figure out where to point AI towards.”

Counterintuitive advice against the standard buy-this-tool content→ IG reel cold open↗ Tweet quote

10:41

“Scaling your business is making yourself less necessary every quarter.”

Aphoristic, standalone, slightly uncomfortable -- high share probability→ newsletter pull-quote↗ Tweet quote

12:15

“The window is open, but it is gonna shut down. In 18 months, it will not be there.”

Urgency close with a specific time horizon→ TikTok hook↗ Tweet quote

The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

analogystory

00:00AI just changed how you can get rich forever. And it's not just buy this tool or try this prompt or sell this AI course. That's kiddie pool stuff.

00:08I've built and scaled AI companies past a million dollars in under six months through Martell Ventures, and I figured out the exact formula on how you can make a lot of fucking money with AI. Starting with stop using AI like a toy. I get it.

00:21AI is fun. We sit there and we're like, look, a new website. Boom.

00:25A new app. Boom. Built a new dashboard.

00:27That's not work. That's just using it like a toy. Rich people put AI to real work, real tasks, real outputs, real problems, big stuff.

00:37AI is selling. It's delivering. It's running their business twenty four seven.

00:41It's not just making them feel good about using AI because it's like outputting things that is like neat to show their friends. It's actually doing work. Like the work you've been avoiding, but it makes you money, that's great work to do with AI.

00:53So how should you use AI? Give AI a real target. Before you buy a single tool, figure out where to point AI towards.

01:02AI makes all the problems no longer problems. So the problem to solve is figuring out what is the problem to solve. And there's always one problem that's way bigger and way more important than the rest, and that is called the bottleneck.

01:14So we gotta point the AI to where you can get the most leverage. That's why I always say, buy back your time. The more time you get back, the more time you can do things that make you more money.

01:23It could reduce waste by having better reporting so you don't make mistakes. It could help you multiply your output. If you know there's a bottleneck around some kind of process that if you unblock it, you can increase your revenue, that is a great place to use AI.

01:36So before you go out there and upgrade your ChatGPT or Clot account to the $200 a month because you've been hearing me talk about ClotCode and you're like, I'm gonna go ClotCode. First, start by figuring out what bottleneck you gotta go solve and then find the right AI solution to solve. It.

01:49So, for example, your Atlas blew up after we featured them in a video. And the founder, Omar, was flooding in calls and bookings and couldn't stay on top of it. So, what did he do?

01:59He didn't go around and buy a bunch of AI tools to like solve this problem in his development or do this other thing in customer success. He looked at the problem, he figured out what the bottleneck was, and he realized that his own tool was the best tool to solve that bottleneck. So how do you find out where you're blocked in your business?

02:13The way I think about it is that every business runs on the same chain. First, have attention. Do people know you exist?

02:19Then you have leads. Okay. They like what you got and they give you their contact info.

02:23Then there's sales, which is a conversation to purchase. Then they buy, so you have to deliver. Then finally, it's retention.

02:28Do they stick around? So now that you understand the chain, here's what you need to do. First, look at each stage of the flow.

02:36Second, we have to ask at every stage, if I 10 x this tomorrow, would my revenue actually go up or would I create another bottleneck? Essentially, in theory, that if you have perfect business, so you got leads going, you have really happy customers on the other side. But at some point, as you're going through each stage, there's probably something that's gonna stop the flow of the customer, the flow of revenue.

02:57For most businesses, they don't have enough attention or they don't have enough leads. Just start there. That's your bottleneck.

03:03Not enough leads means not enough sales, means not enough cash collected, means not enough money to invest in the business. And the truth is is you probably got opportunities and leads that are just falling through the crack.

03:13So where's the money leaking? And look, I built an AI company OS playbook that goes through the whole process in way more detail. It guides you through the audits, the workflow templates, the AI tools I actually use in all my businesses.

03:26If you want it, just DM me YouTube OS on Instagram and I'll send it over. So let's say leads is your biggest bottleneck. You need more revenue.

03:32How do you get AI to solve that? Get AI to generate revenue for you. Starting with leads and sales.

03:39That's where the money enters every business. AI doesn't feel rejection for trying to close a deal. AI doesn't worry about emailing somebody and the person doesn't reply.

03:48Humans get birthed out on sales because they feel the rejection. AI doesn't care. The same effort from you built on the system.

03:55AI executes it 10 x the output. AI can do everything. It can generate leads.

04:00It can automate follow-up and way more. And all you do is close. And that's why in Martell Ventures, we have tools like GetRevio and Your Atlas that we use real AI tools that automate the lead scoring, the qualifying, booking of calls, and way more.

04:13And if you combine those tools with Claude, Manus, you have a whole sales engine where you just come in and close. So how do we get AI to help us find deals and sell? Think of these different levels.

04:24Number one, you have prospecting. AI builds you a lead list. Use tools like Manus, Clogcode, or Apex to scrape socials and find new people every week based on your criteria.

04:34And I'll link all of these up in the description below so you can click them and go sign up for them if that's what you wanna Next, qualifying. AI can filter out all the tire kickers. Yesterday, I was having a conversation with one of my companies because they had a bunch of leads, 20,000 leads sitting there and they're like, I don't wanna talk to everybody.

04:50I wanna talk to the best people that are ready for this product. So we analyze their lead list, did some data enrichment using AI, and found the signals that told us that these people would be perfect for our product.

05:02Like I mentioned, there's tools like your Atlas so they can call qualify. Revio that does this in the social DMs. And what you do is you give it three must haves for each prospect.

05:10It asks those questions to make sure you don't get on the phone with tire kickers. Next, we've got pitching. AI writes hyper personal, very specific messages to your buyers.

05:20In many of my companies, AI will look at past messages to the buyer, look at their current situation, what they're sharing on social media, and write wildly personal messages to each one of them specific to their situation, their industry, and even their stage in the buying process. Next, we have follow-up.

05:38An AI can literally run the whole sequence. If someone doesn't respond, he messages them, email trigger, automatically hyper personalized. How do you do that?

05:46Using AI automation. And finally, upsells. At the end of the day, it is seven times easier to get a current customer to buy more from you than get a person that doesn't know you to buy the first time.

05:57So using AI, you can analyze your customer base and figure out who's bumping up to usage or maybe they're up for renewal and actually tell you and craft a message and send it to them. That's why we build another product called Latch who does this automagically for us. And look, if you wanna get nerdy and get into the specifics and the prompts and the setup, I actually did a whole video going in-depth on all the things that you need to learn how to use a tool called Manus to set this all up.

06:23If you wanna go deep, I'll link it up at the end of this video. So now you've got AI doing the whole sales process. But once you sell, who's doing the delivery that AI just helped you sell?

06:33Here's a cool thing, AI delivers. Most businesses can actually deliver sales, can do really well, but then they drown in delivery and humans hit capacity. And the cool part is AI doesn't have a capacity problem.

06:45It literally can keep doing it. It doesn't sleep. It doesn't get upset.

06:48It doesn't talk back. It's happy all the time. It literally wakes up and is like, I just wanna see you win.

06:53Help me help you win. AI can already do at least 80% of your work. I know, ouch.

06:58How could it possibly do this? This hurts my feelings. I'm such a creative.

07:02It's very important that I'm involved. Let me show you how. So here's a perfect example.

07:06We have an internship program called Young Guns. They hired a bunch of people. They got bottlenecked at onboarding a bunch of Young Guns.

07:13So our default is use AI to solve the problems. So they went to it. It created a first draft of Zero to Hero in fourteen days to teach them how to use AI, but it wasn't perfect.

07:23So they massaged it and then they implemented it and all of a sudden the bottleneck was no longer bottlenecking and got the young guns ready to hit the ground running to act like seasoned pros even though they didn't have the life experience to be ready yet because they were using AI. And that all comes because we continuously train them to default to AI every time.

07:42Don't do anything without first consulting AI because I guarantee it's probably gonna give you a good first draft that you wouldn't know because you don't have the life experience. Now, if you are somebody that has the talent, the knowledge, or an artist, the deep domain experience, I'm gonna show you a way to get out of it but still feel like you have a fingerprint on the creative process.

08:00So the first 10% is what I call the ideation. This is where you open up a chat and you start talking to it about different ideas and opportunities to just even get your bearing straight.

08:10What are you trying to accomplish? What are some concepts? Work with it.

08:13You have expertise. You have ideas. Give it to the AI and have it come back with some other options.

08:18This is where you create the definition of done. What does done look like? The second part is the middle 80%.

08:24This is what's called execution. This is where the AI actually executes the plan, that 10% that you negotiated with it for it to actually help your customers onboard, support them, track them through the process, create a dashboard, create a whole automation, the whole thing.

08:40Let it figure out how and have it check its own work. That's a pro move. Asking the AI to check-in on itself to make sure it's moving towards the right direction based on the first 10%.

08:51The final 10%, this is integration. This is where I like to add my human touch. This is where I review the work.

08:58I add the personal calls. I add the strategic advice. I make sure that things work and feel and look the way I want them to look.

09:05In many ways, the AI is no different than a person that's helping you do the work. You just give it to the AI instead so you don't have to pay them and you don't have to wait. And my favorite thing is waking up every morning to work done for me while I was sleeping.

09:16This is my favorite part, is that 80% done by someone else or by AI is a 100% freaking awesome. It's already giving you a bunch of time back. So now we've got AI selling.

09:27We have AI delivering. But there's one bottleneck left in your business, the most expensive one, and that's you. Now, you want to replace yourself before AI does.

09:37This is the future that I see. There's only two Either you end up working and creating with AI or you're gonna wake up one day and you're gonna be working for it.

09:47So whatever business you have today, if we don't learn how to use the AI to do the work for us, trust me, somebody's gonna come in and they're gonna build an AI and they're gonna take your job. You're gonna feel the same way the newspapers did back when the Internet came. People aren't getting rich by doing more, they're getting rich by doing less.

10:05They're saying, hey, AI, you take that on. I'm gonna take that time to go work on things that only I can do. A few months ago, I decided to shut down my whole company for two days to do an AI hackathon where I put people into teams and encourage them to build tools that would replace their work.

10:21What happened is at the end of those two days, everybody came back 10 times more productive. They built the tooling to replace their work that they didn't want to do in the first place following the ten eighty ten rule, and now using Claude code, CLI at the command line is now mandatory for everybody. Scaling your business is making yourself less necessary every quarter.

10:42So you know you're doing this right as the business grows and you're less needed. Now you'll feel guilty at first. Lean into it.

10:49Allow yourself to breathe for once and have the AI continue to support you in your work. So here's how we make this brain dead simple. I have a process called the camcorder method that I designed originally to hand over things to team members.

11:02But today, I use the same process, but I hand it over to my AI because it's doing the work. So the first thing we have to do is record ourselves doing the task. My favorite way is just get on a Zoom session by yourself.

11:14Share your screen. Record it. While you're talking, you record the screen.

11:17See, the AI can actually take snapshots of what you're looking at so it can build context. And while you're talking, it's taking the transcript so it knows exactly how you're thinking. The whole idea is you want everything recorded or documented in every shape or form.

11:30Two, make sure there is a transcription. So most tools are doing this today. If not, guess what?

11:34The AI can do it for you. Third is feed that transcription, that recording into the AI, and now it has the full context to actually repeat and execute the task that you just talked about. And finally, iterate and refine until AI can do 80% of it.

11:48You'd be surprised how many very complicated steps that AI can do, especially with these multimodals where you can feed it video and images in text. Eventually, it can find the pattern in your process and teach you how to do it even better, but get AI to do that middle 80%. I would suggest doing it at least three meaningful times to give it to the AI to see if it can figure out how you do what you do.

12:10Do this with every task until everything is replaced. I'm talking finance. I'm talking follow-up.

12:15I'm talking hiring. I'm talking all the business stuff, the bottlenecks that would make you more money if they were solved. You need to be replacing yourself.

12:23The window is open, but it's gonna shut down. In eighteen months, it won't be there. So the people who move first won't compete with the rest.

12:31They'll literally lap them. You don't have to be an expert at any stage. You just need to be first.

12:37Right now, it's time to take action and not overthink it. The best skill you can learn is how to unlearn the beliefs and relearn how to activate the AI. I know AI is cool and there's parts of the business that you wish somebody else would take care of, but there's a fear because if it's doing that and you don't know how it's doing that, then maybe it's gonna do it wrong.

12:57I would rather you try and have moments where you might have to get involved and fix something to not even try at all. If we don't create bigger problems, we'll never live a bigger life. How about we go out there and say, no.

13:08Let's try. Let's figure out what breaks. Let's use AI.

13:11Let's be a little crazy in how we think about it. What if it could literally edit a video here? What if it could do a whole sales call for you?

13:20Yesterday, I'm talking to one of my founders and he's telling me how he has his whole sales process automated selling $700 a month subscriptions.

13:28Where he has a whole team that's still taking calls and now he has an AI doing the same calls and eventually the AI is gonna do it better than the team and then he's gonna flip flop them and redeploy those people on separate things. That is a now thing. So either you're like my founder who did it to himself or somebody else is gonna come into your market and just take all of your customers.

13:47So let me know in the comments below. What is the first task you're gonna go and apply AI to that you haven't done yet that you've been putting off that you're like, no, I'm committed to this. I wanna do that.

13:57Comment below and let me know. And remember, if you want my internal AI company OS playbook, just DM me YouTube OS on Instagram and I'll send it right over. And as promised, here's a video that'll teach you how to get AI to take over your whole sales pipe

The Hook

The bait, then the rug-pull.

The opener does not hedge. Before any framework or credential, the first sentence makes an absolute claim about wealth and permanence -- a bet that curiosity or skepticism will hold viewers through the next 14 minutes either way.

Frameworks